Hi, I wanted to try the new 3.0 Linux Kernel, and I got stucked for a freeze at the start of X.
I have bisected the problem and the error was on commit a51f7a6, wich enable rc6 by default (some time ago I have sent another bug about that).
I have reverted the commit (the complete patch is attached) and now everything works great.
Now, I have a Lenovo X201s, and if you need more information for debugging purpose just let me know, otherwise just apply the patch attached (on top of -rc4).
If you prefer I can send the patch using git-send-email.
Thanks
Bye Francesco
On Tue, 21 Jun 2011 16:21:33 +0200, Francesco Allertsen fallertsen@gmail.com wrote: Non-text part: multipart/mixed
Now, I have a Lenovo X201s, and if you need more information for debugging purpose just let me know, otherwise just apply the patch attached (on top of -rc4).
Given the number of X201s machines running this kernel (like the one I'm using right now), I'd love to get this fixed for you instead of disabling rc6 -- rc6 is worth a ton of power savings.
Please describe your configuration (which OS, which X and i915 driver versions) and then send along a kernel log with drm.debug=0x04 so we can see what's going on. If you stick these in a bugzilla entry, we'll not lose track of them.
On Tue, Jun 21, 2011 at 08:28:17AM -0700, Keith Packard wrote:
Given the number of X201s machines running this kernel (like the one I'm using right now), I'd love to get this fixed for you instead of disabling rc6 -- rc6 is worth a ton of power savings.
I didn't know that, anyway, I'm don't know the intel driver so well to solve the problem, that's why I did only the revert, and I'll give you more information to solve the issue.
Please describe your configuration (which OS, which X and i915 driver versions) and then send along a kernel log with drm.debug=0x04 so we can see what's going on. If you stick these in a bugzilla entry, we'll not lose track of them.
Let's start.
I'm using Slackware -current with the xorg-server version 1.9.5 and the kernel 3.0.0-rc4 (with the revert applied).
Attached there is the full dmesg output with drm.debug=0x04.
If you prefer that I open a bug on the kernel bugzilla to put all the informations no problem.
Bye Francesco
On Tue, 21 Jun 2011 17:52:05 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
If you prefer that I open a bug on the kernel bugzilla to put all the informations no problem.
I'm more interested in seeing how the dmesg output differs between the working state (rc6=0) and the broken state (rc6=1). Or is the machine not accessible over ssh once you've started X?
On Tue, Jun 21, 2011 at 09:11:29AM -0700, Keith Packard wrote:
I'm more interested in seeing how the dmesg output differs between the working state (rc6=0) and the broken state (rc6=1). Or is the machine not accessible over ssh once you've started X?
I got a complete freeze when I start X. I have tried to log the output of dmesg via ethernet using netconole, but it doesn't log the drm debug. Is there another parameter to append at the kernel to send via ethernet also the debug?
Bye Francesco
On Tue, 21 Jun 2011 18:42:09 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
On Tue, Jun 21, 2011 at 09:11:29AM -0700, Keith Packard wrote:
I'm more interested in seeing how the dmesg output differs between the working state (rc6=0) and the broken state (rc6=1). Or is the machine not accessible over ssh once you've started X?
I got a complete freeze when I start X. I have tried to log the output of dmesg via ethernet using netconole, but it doesn't log the drm debug. Is there another parameter to append at the kernel to send via ethernet also the debug?
I don't know how to get the drm debug sent to ethernet; perhaps someone else does?
Btw, can you send along the output of dmidecode? I'm wondering what version of the BIOS you've got. We've seen some pretty strange behaviour on X201s with earlier BIOS versions which have been fixed with newer versions.
For comparison, my X201s says:
# dmidecode 2.9 SMBIOS 2.6 present. 78 structures occupying 2868 bytes. Table at 0x000E0010.
Handle 0x0000, DMI type 0, 24 bytes BIOS Information Vendor: LENOVO Version: 6QET46V1 (1.16 ) Release Date: 07/05/2010 Address: 0xE0000 Runtime Size: 128 kB ROM Size: 8192 kB Characteristics: PCI is supported PC Card (PCMCIA) is supported PNP is supported BIOS is upgradeable BIOS shadowing is allowed ESCD support is available Boot from CD is supported Selectable boot is supported EDD is supported 3.5"/720 KB floppy services are supported (int 13h) Print screen service is supported (int 5h) 8042 keyboard services are supported (int 9h) Serial services are supported (int 14h) Printer services are supported (int 17h) CGA/mono video services are supported (int 10h) ACPI is supported USB legacy is supported BIOS boot specification is supported Targeted content distribution is supported BIOS Revision: 1.22 Firmware Revision: 1.9
On Tue, 21 Jun 2011 11:04:15 -0700 Keith Packard keithp@keithp.com wrote:
On Tue, 21 Jun 2011 18:42:09 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
On Tue, Jun 21, 2011 at 09:11:29AM -0700, Keith Packard wrote:
I'm more interested in seeing how the dmesg output differs between the working state (rc6=0) and the broken state (rc6=1). Or is the machine not accessible over ssh once you've started X?
I got a complete freeze when I start X. I have tried to log the output of dmesg via ethernet using netconole, but it doesn't log the drm debug. Is there another parameter to append at the kernel to send via ethernet also the debug?
I don't know how to get the drm debug sent to ethernet; perhaps someone else does?
Use netconsole. Documentation/networking/netconsole.txt has all the gory details.
On Tue, Jun 21, 2011 at 11:29:41AM -0700, Jesse Barnes wrote:
Use netconsole. Documentation/networking/netconsole.txt has all the gory details.
I have used netconsole, with that documentation, but it doesn't show me the drm debug messages, only the other dmesg messages, and I don't know why.
Bye Francesco
On Tue, Jun 21, 2011 at 08:48:55PM +0200, Francesco Allertsen wrote:
On Tue, Jun 21, 2011 at 11:29:41AM -0700, Jesse Barnes wrote:
Use netconsole. Documentation/networking/netconsole.txt has all the gory details.
I have used netconsole, with that documentation, but it doesn't show me the drm debug messages, only the other dmesg messages, and I don't know why.
Try to run dmesg -n 8 before loading the module.
Marcin
On Tue, Jun 21, 2011 at 09:12:48PM +0200, Marcin Slusarz wrote:
Try to run dmesg -n 8 before loading the module.
I'm using the netconsole directly at the boot, because I got the freeze, so I don't know how to pass that value directly appended to the kernel (if it's possibile).
If you want I can boot up with the initlevel 3 so I don't have the freeze, and then start netconsole and start X.
Bye Francesco
On Tue, Jun 21, 2011 at 11:04:15AM -0700, Keith Packard wrote:
Btw, can you send along the output of dmidecode? I'm wondering what version of the BIOS you've got. We've seen some pretty strange behaviour on X201s with earlier BIOS versions which have been fixed with newer versions.
Here is mine:
# dmidecode 2.11 SMBIOS 2.6 present. 78 structures occupying 2864 bytes. Table at 0x000E0010.
Handle 0x0000, DMI type 0, 24 bytes BIOS Information Vendor: LENOVO Version: 6QET52WW (1.22 ) Release Date: 08/23/2010 Address: 0xE0000 Runtime Size: 128 kB ROM Size: 8192 kB Characteristics: PCI is supported PC Card (PCMCIA) is supported PNP is supported BIOS is upgradeable BIOS shadowing is allowed ESCD support is available Boot from CD is supported Selectable boot is supported EDD is supported 3.5"/720 kB floppy services are supported (int 13h) Print screen service is supported (int 5h) 8042 keyboard services are supported (int 9h) Serial services are supported (int 14h) Printer services are supported (int 17h) CGA/mono video services are supported (int 10h) ACPI is supported USB legacy is supported BIOS boot specification is supported Targeted content distribution is supported BIOS Revision: 1.34 Firmware Revision: 1.5
Bye Francesco
On Tue, 21 Jun 2011 20:47:57 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
Version: 6QET52WW (1.22 ) Release Date: 08/23/2010
Thanks. I'll see if I can't get an X201s upgraded to this version and see if we can reproduce the problem you're having...
On Tue, Jun 21, 2011 at 12:01:34PM -0700, Keith Packard wrote:
Thanks. I'll see if I can't get an X201s upgraded to this version and see if we can reproduce the problem you're having...
Any news?
I have found the time today for more debugging.
Attached there are 3 files.
dmesg-startup is the startup of the last 3.0.0-rc6 kernel, with the debug enabled. dmesg-enabled is the X startup (and freeze) with debug mode and rc6 enabled. dmesg-disabled is the X startup with debug mode enabled and rc6 disabled (so no freeze; full kde startup and logout).
I hope this helps you to solve the problem, otherwise the X201s users will not use the 3.0.0 kernel :-(.
If you need more informations just let me know.
Bye Francesco
On Wed, Jul 6, 2011 at 2:01 AM, Francesco Allertsen fallertsen@gmail.com wrote:
On Tue, Jun 21, 2011 at 12:01:34PM -0700, Keith Packard wrote:
Thanks. I'll see if I can't get an X201s upgraded to this version and see if we can reproduce the problem you're having...
Any news?
I have found the time today for more debugging.
Attached there are 3 files.
dmesg-startup is the startup of the last 3.0.0-rc6 kernel, with the debug enabled. dmesg-enabled is the X startup (and freeze) with debug mode and rc6 enabled. dmesg-disabled is the X startup with debug mode enabled and rc6 disabled (so no freeze; full kde startup and logout).
I hope this helps you to solve the problem, otherwise the X201s users will not use the 3.0.0 kernel :-(.
If you need more informations just let me know.
AFAICT, this problem is still there in 3.0-rc7. Keith, isn't it time to revert commit a51f7a6 ("drm/i915: enable rc6 by default") before 3.0 is out?
On Tue, 12 Jul 2011 10:33:06 +0300, Pekka Enberg penberg@kernel.org wrote:
AFAICT, this problem is still there in 3.0-rc7. Keith, isn't it time to revert commit a51f7a6 ("drm/i915: enable rc6 by default") before 3.0 is out?
Yes. I'll put together a few tiny patches including this.
On Tue, 12 Jul 2011 10:33:06 +0300, Pekka Enberg penberg@kernel.org wrote:
AFAICT, this problem is still there in 3.0-rc7. Keith, isn't it time to revert commit a51f7a6 ("drm/i915: enable rc6 by default") before 3.0 is out?
On Tue, Jul 12, 2011 at 6:42 PM, Keith Packard keithp@keithp.com wrote:
Yes. I'll put together a few tiny patches including this.
This if fixed in mainline as of commit a94919eaddaa3fede1df8563ce4d761a75374645 ("Revert "drm/i915: enable rc6 by default"). Keith, it would have been nice to include a 'Reported-by' and/or a link to this discussion in the changelog for future reference.
Pekka
On Thu, Jul 14, 2011 at 10:37:15AM +0300, Pekka Enberg wrote:
This if fixed in mainline as of commit a94919eaddaa3fede1df8563ce4d761a75374645 ("Revert "drm/i915: enable rc6 by default"). Keith, it would have been nice to include a 'Reported-by' and/or a link to this discussion in the changelog for future reference.
The correct commit is 05bd42688dbc066d4e2689b6f73c0470601f788b, the one you mentioned is "Idling requires waiting for the ring to be empty".
Bye Francesco
On Thu, 14 Jul 2011 09:50:07 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
The correct commit is 05bd42688dbc066d4e2689b6f73c0470601f788b, the one you mentioned is "Idling requires waiting for the ring to be empty".
We mentioned the 'Idling' fix as it may make RC6 work. Have you tried that with RC6 enabled on the command line (or with 05bd42688dbc066d4e2689b6f73c0470601f788b not applied)? Obviously, not enabling RC6 by default will 'fix' things, but only with a huge hit to power consumption.
On Thu, Jul 14, 2011 at 09:06:20AM -0700, Keith Packard wrote:
We mentioned the 'Idling' fix as it may make RC6 work. Have you tried that with RC6 enabled on the command line (or with 05bd42688dbc066d4e2689b6f73c0470601f788b not applied)? Obviously, not enabling RC6 by default will 'fix' things, but only with a huge hit to power consumption.
I have tried to boot with the latest git with the commit 05bd42688dbc066d4e2689b6f73c0470601f788b reverted (so I have the 'Idling fix' and the rc6 enabled), but I have the same freeze.
Bye Francesco
On Thu, 14 Jul 2011 19:00:26 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
I have tried to boot with the latest git with the commit 05bd42688dbc066d4e2689b6f73c0470601f788b reverted (so I have the 'Idling fix' and the rc6 enabled), but I have the same freeze.
Thanks.
On Thu, 14 Jul 2011 19:00:26 +0200, Francesco Allertsen fallertsen@gmail.com wrote:
I have tried to boot with the latest git with the commit 05bd42688dbc066d4e2689b6f73c0470601f788b reverted (so I have the 'Idling fix' and the rc6 enabled), but I have the same freeze.
Can you send me your kernel .config file? I'm still not having any luck reproducing your problems, using an X201s with the 1.22 BIOS version.
I'm wondering if the trouble is caused by the the selection of kernel config parameters....
On Fri, Aug 12, 2011 at 12:53:52PM -0700, Keith Packard wrote:
Can you send me your kernel .config file? I'm still not having any luck reproducing your problems, using an X201s with the 1.22 BIOS version.
I'm wondering if the trouble is caused by the the selection of kernel config parameters....
Kernel config attached.
Bye Francesco
On Thu, 14 Jul 2011 10:37:15 +0300, Pekka Enberg penberg@kernel.org wrote:
This if fixed in mainline as of commit a94919eaddaa3fede1df8563ce4d761a75374645 ("Revert "drm/i915: enable rc6 by default"). Keith, it would have been nice to include a 'Reported-by' and/or a link to this discussion in the changelog for future reference.
Yes, I'm sorry I didn't remember to include a link.
dri-devel@lists.freedesktop.org
-
Francesco Allertsen -
Jesse Barnes -
Keith Packard -
Marcin Slusarz -
Pekka Enberg